`Javier Bezos' wrote > > >> On the other hand using different encodings for a language within one > >> document is somewhat perverse. So it would be reasonable to have per > >> document defaults. > > > >Yes, but these defaults should depend on the language. > > No - they depends on the system. A Mac file is the same in Spain > and France. A Linux file is the same in Spain and France. But a > Spanish file is not the same in Mac and Linux. Actually you misunderstood me, because I meant "depend on the language" and not "depend only on the language." Of course, there can be other parameters, like platform, etc. So, we could decide a default platform (:-) or make the platform explicit. I do not see why "A Linux file is the same in Spain and France" should be always true (assuming you mean that such a file is viewed the same way), because this can depend on the linux configuration, or on the configuration of your editor. I have a colleague who has a Russian configuration on his UNIX box and he doesn't see files as I do. When you say "Spanish file is not the same in Mac and Linux," you mean presumably that the keyboard configurations and editor (or other display programs) are different, but a file is a file! So, even if there are several kinds of dependences, I still think that there should be default encodings per language, and if you wish, per platform (the latter information could be obtained from the inputenc package, or from a global option for instance). One thing that should also be clear, is the fact that one can manipulate text in a given encoding, without actually having a configuration corresponding to this encoding. For instance, I have manipulated Russian text in koi8-r encoding, without actually seeing the characters properly other than with xdvi. In this case, I really didn't care about what I saw, nor did I care on the platform. I only fed the Russian environment with koi8-r text, and if koi8-r had been the default input encoding, this would have resulted in no hassle at all. Denis Roegel